In this folder you'll find the dataset and code to reproduce all analysis in both the paper and the online appendix.

All analysis is done in R (v. 3.5.1), in addition to base R the following packages are needed to reproduce all files:
 - 'tidyverse'
 - 'list'
 - 'xtable'
 - 'rio'
 - 'texreg'

The code is annotated to improve readability. To reproduce tables and figure please.

1. run **main-analysis.R** (which calls main-estimates.R) to produce main estimates and output for *figure 2*, *table 7* and *table 8*

2. run **subgroup-analysis.R** (which calls subgroup-system.R, subgroup-censor.R, subgroup-corruption.R and subgroup-confidence.R,) to produce estimates and output for *figure 4*

3. run **no-design-effect** to produce *table 5* and *6*.
